Repository navigation
Conversation
|
Please provide some examples of errors with the new code The TryCatch part applies to lot of places |
@rom1504 these are not useful, however if it could also log the packet at the same time, the issue would become very obvious: please consider this, IMO its better rather than not providing any additional info (right now there is no way to get the packet from the exception at all) |
rom1504
left a comment
There was a problem hiding this comment.
Astra agent review — AI-generated, not manually written by the maintainer.
The examples now explain why additional diagnostics help, so the earlier request for examples is answered. The remaining decision is scope: tryCatch is shared by codecs, and changing global Error.stackTraceLimit around every invocation affects the whole synchronous call tree; unconditional packet serialization also changes error-message size and exposes complete packet contents to logging.
Prefer an explicit diagnostic option or a narrowly owned error path, with bounded packet context. Coordinate actual compiled field paths with #180 so useful location information does not depend on logging an entire value. This is a design/maintenance assessment, not a measured performance regression; no benchmark or fresh suite was run.
Skills used: prismarine-review checked the unchanged diff and current discussion; prismarine-architecture-review traced the public API, ownership and failure behavior.
Some error reporting tools like Sentry already do increase stack trace limit, but I think it would be good to make this default behavior. example stack trace with the extended limit:
If you don't like the idea of increasing the stack trace consider at least adding the packet to the message that caused the problem